<!-- This model card has been generated automatically according to the information the Trainer had access to. You should probably proofread and complete it, then remove this comment. -->
wav2vec2-xls-r-300m-asr_xh-run3
This model is a fine-tuned version of facebook/wav2vec2-xls-r-300m on an unknown dataset. It achieves the following results on the evaluation set:
- Loss: 0.4859
- Wer: 0.4817
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 0.0003
- train_batch_size: 4
- eval_batch_size: 4
- seed: 42
- gradient_accumulation_steps: 3
- total_train_batch_size: 12
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: linear
- lr_scheduler_warmup_steps: 500
- num_epochs: 30
- mixed_precision_training: Native AMP
Training results
Training Loss | Epoch | Step | Validation Loss | Wer |
---|---|---|---|---|
3.0804 | 1.2 | 250 | 3.0878 | 1.0 |
0.8823 | 2.39 | 500 | 0.7691 | 0.8714 |
0.521 | 3.59 | 750 | 0.4781 | 0.6521 |
0.3886 | 4.78 | 1000 | 0.4321 | 0.6152 |
0.2966 | 5.98 | 1250 | 0.4262 | 0.5983 |
0.2148 | 7.18 | 1500 | 0.4399 | 0.5410 |
0.1843 | 8.37 | 1750 | 0.4483 | 0.5219 |
0.1491 | 9.57 | 2000 | 0.4190 | 0.5011 |
0.1348 | 10.77 | 2250 | 0.4657 | 0.5154 |
0.1247 | 11.96 | 2500 | 0.4986 | 0.5112 |
0.1015 | 13.16 | 2750 | 0.4859 | 0.4817 |
Framework versions
- Transformers 4.28.0
- Pytorch 2.0.1+cu117
- Datasets 2.14.4
- Tokenizers 0.13.3